AN EMPIRICAL-INVESTIGATION OF RISK CLASSES - ARE COMMON PROXIES VALID

This article constructs observed corporate risk classes and evaluates the accuracy of commonly used industry and broader-than-industry groups as proxies. Observed risk classes, both systematic and idiosyncratic, are obtained using cluster analysis to group firms according to their risk profiles, determined with a simple LFM. Comparison tests between the common proxies and the observed risk classes show that neither the industry nor the broader proxies correspond to systematic risk classes, but that certain broader-than-industry groups may provide reasonable approximations for idiosyncratic risk classes. The analysis of the observed risk classes indicates that industry specific effects are inherent in both idiosyncratic and systematic risk.